EXPERIMENTAL
NEUROLOGY
Physiological Longus
54, 478-488 (1977)
Alterations Motor Nerve of Surgical
of Rat Extensor Digitorum Terminals as a Result Denervation
DIANA J. CARD 1 UnizIersity
of
Califontia,
Department Received
of Biology, July
Los Angeles,
California 90024
30,1976
The sequence of presynaptic degeneration changes which occur as a result of denervation of rat extensor digitorum longus muscle was studied. Reduction of endplate potential quanta1 content was the first neuromuscular alteration. Subsequently, endplate potentials disappeared although spontaneous transmitter release continued, at a higher frequency than in control muscles. During repetitive stimulation of the nerve, partial conduction block of the nerve terminal occurred at stimulation frequencies as low as I/s. Few terminals were able to respond at frequencies greater than 5/s. Transmitter release was quanta1 ; however, responses at synapses with a quanta1 content between one and five did not fit a Poisson distribution. The hypothesis is presented that, as the nerve terminal progressively degenerates after nerve transection, a few sites of transmitter release often persist for several hours, and these sites have a probability of release equal to or higher than normal.
INTRODUCTION Mammalian neuromuscular synapses are structurally and functionally changed as a result of denervation (1, 7, 10, 16-18). The first presynaptic alterations, resulting in increased miniature endplate potential (mepp) frequency several hours before spontaneous transmitter release disappears (1, 7, 18)) might be the result of depolarization of the nerve terminals (9) or they might involve the disruption of terminals and their engulfment by Schwann cells (16-18). These changescould be accompanied by alterations in coupling between invading presynaptic nerve action potentials and trans1 I would like to thank Dr. Alan Grinnell for his helpful advice and guidance throughout these experiments. This research was supported by USPHS Grant NS06232 to Dr. Grinnell and a predoctoral fellowship on USPHS Grant GM 448 to the author whose present address is Carnegie Institution of Washington, Department of Embryology, 115 West University Parkway, Baltimore, Maryland 21210. 478 Copyright All rights
@ 1977 by Academic Press, of reproduction in any form
Inc. reserved.
ISSN
0014-4886
MOTOR
NERVE
TERMINALS
AND
DENERVATION
479
mitter releaseor by failure of nerve action potentials to invade the terminals and could provide insight into the events occurring tluring degeneration. The purpose of the following experiments was to determine the sequence of presynaptic degenerative changes which occur as a result of nerve lesion. Other investigators have already documented postsynaptic alterations (1. 7, 10, 16-1s). The efficacy of evoked and spontaneous transmitter release was determined at different times after denervation. The occurrence of subthreshold endplate potentials, which is unusual in the absence of pre- or postsynaptic blocking agents, was then examined in detail. MATERIALS
AND METHODS
Extensor digitorum longus muscles from male Wistar rats (150 to 250 g) were denervated under light ether anesthesia using aseptic techniques. Surgical procedures and the methods for removal of the muscle, intracellular recording, and locating the neuromuscular junction were described previously (7). The contralateral extensor digitorum longus muscle served as a control: Its nerves were carefully separated from surrounding tissuesbut were not crushed or cut. Newowmm~lar Jmctio~ Recording. Spontaneous miniature endplate potentials with rise times of less than 1 ms were used as the criterion for locating neuromuscular junctions. They were observed for several minutes and either photographed on an oscilloscope screen or recorded with a Clevite-Brush Mark 220 pen recorder. Then the presence or absence of neuromuscular transmission at the synapse was tested by stimulating the entire extensor digitorum longus nerve stump using a suction electrode and recording the muscle fiber’s response. No attempt was made to reduce muscle twitching or endplate potential amplitude with curare or presynaptic blocking agents. The muscles were bathed in warmed (25 to 26”C), continuously perfused, oxygenated (95% 02, 5% CO,) bicarbonate-buffered saline (7) and remained in good condition for 8 h. No further denervation changes were observed after the muscles were placed in the bath; that is, synapsesat the same distance from the nerve cut gave similar responsesat the beginning and at the conclusion of the recording session. RESULTS Action Potentials. Until about 9 h after denervation, at endplates 1 mm from the nerve lesion, the only postsynaptic response to nerve stimulation was a muscle action potential (Table 1). The muscle twitched vigorously in responseto nerve activation. Muscle fibers in the control extensor digitorum longus always contracted in responseto nerve stimulation, even after 8 h in the bath.
480
DIANA
J.
TABLE Percentage
Distance (mm)
6-7
1
of Fibers in Denervated Extensor Digitorum Longus Exhibiting Action Potential, Endplate Potential, or No Responsea Denervation time (h)
1
CARD
Number fibers
Action potentials (%I
5 7 9 10 12 14
8.5 37 43 25 27 3
98 97 30 28 37 -
7 9 10 12 14 16 18 20
17 32 22 19 13 12 3 25
100 90 82 74 77 50 33 -
a All measurements were made a rise time of less than 1 ms.
from
fibers
having
Subthreshold endplate potentials (%I
an
No response (73
1
1 3 44 60 52 100
26 12 11 -
-
10
4 5 8 17 -
14 21 1.5 33 67 100
miniature
endplate
potentials
with
Subthreshold Endplate Potentials. Between 9 and 12 h after denervation, subthreshold endplate potentials were recorded from 5 to 20% of the synapses that had miniature endplate potentials (Fig. 1). The mean quanta1 content, WZ=
mean endplate potential amplitude mean miniature endplate potential amplitude’
Cl1
at these denervated neuromuscular junctions was greatly reduced compared to the response of the contralateral control and of denervated synapses prior to this time. Intermediate-size endplate potentials were observed whose amplitude varied in response to the same stimulus (Fig. la). At later denervation times the decrease in quanta1 content became so severe that small endplate potentials, with a quanta1 content between one and four, were often seen (Fig. lb). Four examples of this are summarized in Table 2. In these fibers, the endplate potential was usually the size of a single or a double miniature endplate potential, as opposed to the 100 to 200 quanta that comprise an endplate potential in normal muscle (5, 8). Subthreshold endplate potentials were never seen in contralateral control extensor digitorum longus muscles.
MOTOR
NERVE
TERMINALS
AiXD
481
DENERVATION
1mV
L
10 msec
2mV J 5msec
FIG. 1. a-Subthreshold endplate potential recorded from a synapse less than 1 mm from the cut nerve 8 h after denervation. Quanta1 content of this fiber was 17.5 ; miniature endplate potential frequency was 2.5/s. b-Low quanta1 endplate potential recorded at a synapse less than 1 mm from the nerve cut 8 h after denervation. Resting membrane potential was 83 mV. Quanta1 content was 1.2 ; miniature endplate potential frequency was 28.3/s.
After a 7-h denervation, the proportion of muscle fibers 1 mm from the nerve lesion with action potentials decreased, while the proportion of fibers with no response steadily increased. The percentage of synapseswith subthreshold endplate potentials was low but remained approximately conTABLE Quanta1
Fiber
A
Content of Four Different Extensor Digitorum
Resting potential
of failures
Actual
Predicted
84 111
0 46
31s
47 32 43
70 mV 1 /s 2/s
21-3
B
80 mv l/S
2/s
C
n
78 mv lis
21s ~1Measured
by Eqs. Cl] and
of
Muscle Fibers Longus Muscles”
Number
79 mv l/S
Number stimuli
2 from
Eq. Cl1
Eq. C31
Miniature endplate potentials per second
16 47
1.65 0.86
0.86
2.7
15 19 43
8 18
1.77 0.59 0
1.12 0.52
5.2
42 83
3 7
5 22
2.09 1.32
3.36 3.02
0.5
50 28
2 13
19 19
0.9i 0.39
3.22 0.77
3.8
[.~I at different
stimulation
frequencies.
Quanta1
Denervated
content
482
DIANA
J.
CARD
stant while synaptic transmission was present (Table 1). The synapses 6 to 7 mm from the nerve lesion showed the same pattern of deterioration of response, i.e., action potential, endplate potential, no response, as those at 1 mm ; however, the onset of each step occurred later (Table 1). The number of failures of subthreshold endplate potential response at low-frequency stimulation (l/s) in most cases, was fewer than predicted by Poisson’s law according to the quanta1 hypothesis. If x is the number of quanta in any one response, m is the mean of all values of x (the average of all responses of the number of quanta per response), and N is the total number of responses, then the frequency of endplate potentials, no/N, R/N, n2/N, in the classes x = 0, 1, 2, . . . can be determined by the Poisson distribution :
m” -nz = -pn* N x! The number of failures (when be predicted by the expression:
x = 0) at a given mean response
no/N
= ecm.
(+a) can
iI31
The average response (~2) was calculated by Eq. [ 11. At a synapse with quanta1 release, both ways of determining m, Eqs. [l] and [3] describe the mean quanta1 response (9). Therefore, 1n-N = n0
mean endplate potential amplitude mean miniature endplate potential amplitude’
c41
Table 2 shows the results of predictions of Poisson’s law compared with the actual number of failures when the nerve was stimulated. The actual number of failures did not agree with that predicted by Poisson. At l/s stimulation, the failure rate is lower than expected, except for fiber B. If the probability of quanta1 release is low (p approaches 0), then the frequency of endplate potential responses of x = 0, 1, 2, . . . approximates the Poisson distribution, and a large number of failures is expected. Because the number of failures of response at the degenerating synapses was lower than expected for the observed endplate potential amplitudes, the probability of transmitter release was estimated. Two binomial expressions : m = (1 - p)/v’ (where
ZJ= coefficient of variation
of endplate potential
-P m = In (1 - p)
lnN no’
responses)
and
MOTOR
NERVE
TERMINALS
AND
483
DENERVATION
were cc~rrhinrtl to give 1 P 2 = (1 - p) In (1 - p) ln?+ The quantity
l/v’
was plotted versus In no/N and the slope of the graph,
P (1 - P>In (1 - PI’ was calculated to give an estimate of p [Fig. 9 in ( 15) 1. At the degenerating extensor digitorum longus nerve terminals, p was estimated to be approximately 0.8 compared to that measured by other investigators in curarized rat diaphragm, about 0.2 (8). When the nerve was stimulated at frequencies above l/s, the number of failures increased, so that in some cases the actual and predicted number of failures was the same (Table 2, Fig. 2, 3). In Fig. 2A, the actual response amplitude remained the same, but the number of zero responses increased. In the muscle fiber of Fig. 2B, the endplate potential response size decreased along with an increased number of failures at higher stimulation frequencies. Figure 3 illustrates both reduced endplate potential size and failure of response in a muscle fiber stimulated at rates from one stimulus every 2 s to 10/s. Each graph represents the sequence of responses of the muscle fiber (endplate potentials) at the designated frequency. The muscle was allowed to rest for 1 min between each set of stimuli. Failure (no response) first 6.
J/set
mu 6AC6 o‘e ::I2 0
0.2
04
06
mV
0.8
ID
mV
FIG. 2. Endplate and miniature endplate potential (mepp) amplitude histograms. Endplate potentials were measured at a stimulation rate of 1, 2, or 3/s. These graphs illustrate that release is probably quanta1 and consists of only one, two, or three units. Increasing the stimulation rate increases the number of response failures. A-Fiber A of Table 2; B-Fiber B.
484
DIANA
J.
CARD
I/ 2sec
,;)&
io” ’ 40’“’ 63 ’ do* ’ IASsec
S/set
41JJ!Lec sequence
I om 20’” of responses
FIG. 3. Endplate potential size in millivolts of the lst, 2nd, 3rd, . . ., 20th, . . ., 40th, . . ., 60th, . . . etc. response when the nerve is stimulated every 2, 1.5, 1 s, etc. The muscle was allowed to rest 1 min between each stimulus regime.
appeared at a stimulation rate of 2/s. With increasing frequency of stimulation the number of failures also increased. Additionally, the number of full responsesat the beginning of the stimulation varied little with increasing stimulation frequency : A stepwise decrease of endplate potential amplitude occurred after about 10 stimuli at all frequencies. Because the responsewas decreased to less than half from stimulus 10 to 11 (at l/s), it is possible that the amount of nerve terminal releasing transmitter was similarly reduced after 10 stimuli. In fact, the initial “full” response of this terminal probably represents a vastly reduced releasearea compared to normal junctions. Some muscle fibers that spiked in responseto nerve stimulation occasionally failed to respond even with a subthreshold endplate potential at stimulation frequencies of 5 or 10/s, as if conduction block occurred at a branch point of the motor unit before the nerve terminal.
MOTOR
NERVE
TERMIXALS
AND
485
DENERVATION
0.p. L..
.
c zo:
3z
I ““I ePP
iotJ
f
II
A-“,.,,..,,..,.,...., zo-
no ePP
0
I” IO
7 mepps
I
20
r
I
30
T 1-1
40
/second
FIG. 4. Miniature endplate potential frequency versus frequency of occurrence in muscle fibers with a suprathreshold response eliciting action potentials (a.p.), subthreshold endplate potentials (epp) , and no response to indirect stimulation (no epp) . There are more fibers with higher frequencies in the “no epp.” category; however, the majority of values in each case is similar.
In summary, just prior to failure of synaptic transmission, the degenerating terminal was unable to follow stimuli at or greater than 2/s, resulting in depression of the endplate potential and possibly conduction block of the nerve action potential. The low quanta1 endplate potentials probably result from a reduced area of nerve terminal which releasestransmitter at a probability equal to or greater than normal. Disappearance of Endplate Potentials. Evoked release started disappearing about 9 h after denervation, although spontaneous transmitter release was still present at those synapses. Kot all fibers in the same region of the muscle became nonfunctional simultaneously. About the time evoked release disappeared, miniature endplate potential frequency increased (7). Then miniature endplate potentials disappeared and no synaptic response could be measured in the muscle. An attempt was made to correlate miniature endplate potential frequency with efficiency of evoked transmission. Figure 4 shows that fibers with no endplate potential have a broader range of miniature endplate potential frequencies than fibers with a subthreshold endplate potential or an action potential. However, most of the fibers measured in all groups have similar miniature endplate potential frequencies. DISCUSSION The sequence of presynaptic degenerative changes occurring at the extensor digitorum longus neuromuscular junction was : (i) reduced transmitter output and a subthreshold endplate potential, (ii) inability of the terminal to evoke synchronous transmitter release, although spontaneous miniature endplate potentials were still present or even increased in frequency, and (iii) disappearance of nliniature endplate potentials.
486
DIANA
J.
CARD
A permanent reduction of functional nerve terminal area as a result of denervation could account for the decrease in endplate potential quanta1 content observed after 9 h of denervation. As the nerve terminal degenerates, the amount of functional membrane may be progressively reduced so that the response size dwindles to nothing as fewer and fewer sites of transmitter release remain. Many degenerating axon terminals contain swollen and disrupted mitochondria (1618). Mitochondrial dysfunction could produce bursts of transmitter release (2, 12) and diminish or eliminate endplate potentials. As the nerve terminals become fragmented and engulfed by Schwann cells (17, 18) all synaptic activity disappears (17). Accordingly, when degenerating terminals are fixed in glutaraldehyde immediately after miniature endplate potentials disappeared, one sees numerous partially engulfed terminal axons lacking synaptic vesicles ( 18). Branch point blockade in the motor axon especially at high stimulation frequencies could explain the stepwise decline in response during a train. This would represent a reversible decrease in the area of terminal that is capable of releasing transmitter. After the spike was blocked in one portion of the nerve terminal, the other(s) continued to release quanta in a fluctuating manner, characteristic of endplate potentials recorded from Mg-blocked preparations. Complete failure of response could be due to conduction block proximal to the endplate, such as at a branch of the motor neuron to another muscle fiber of the motor unit. Normal mammalian neuromuscular junctions do exhibit conduction block, but the frequencies necessary to produce block are 20 to 40/s rather than the 2 to 5/s seen in this study (13, 14). However, the fact that the denervated nerve terminal is degenerating could probably account for the reduced conduction at branch points. With stimulation frequencies above l/s, the actual number of complete failures of response, in some cases, agrees with that predicted by Poisson statistics. The probability of release is probably decreased in these cases by failure of the action potential to invade the nerve terminals. Therefore, the similarity of the actual number of failures and those predicted by Poisson statistics may be coincidental. Subthreshold endplate potentials have been reported at regenerating neuromuscular junctions (3, 4). The sequence of establishment of transmission at the regenerating synapses is the reverse of that seen at degenerating endplates in this study: Miniature endplate potentials are usually seen before endplate potentials, then subthreshold endplate potentials occur, and finally muscle action potentials occur in response to nerve stimulation (3, 4). Subthreshold endplate potentials are also found in muscles partially blocked by botulinum toxin (6) and at neuromuscular junctions of mice affected by “motor endplate disease” (I 1). In each of these cases, the area of terminal capable of evoking synchronous transmitter release is probably
MOTOR
KERVE
TERMINALS
AND
DENERVATION
487
reduced. In particular, at synapses of the “motor endplate disease” mutant, action potentials fail to invade nerve terminals (11). Subthreshold endplate potentials have not been seen in degenerating rat diaphragm (17). Transmission loss occurs abruptly ; nerve stimulation either produces a muscle action potential or no response. The responses seen in the present study that have extremely low quanta1 content (e.g., wz = 2), but also a low failure rate, are unusual and are difficult to explain purely on the basis of the quanta1 hypothesis. Normally, for quanta1 analysis, increased magnesium and reduced calcium concentrations are used to decrease the probability of a particular quanta1 unit escaping from the nerve terminal (9). At normal neuromuscular junctions there is a relatively high probability of transmitter release and a correspondingly large surface area of functional terminal axon. One interpretation of the present data is that at this stage of degeneration only a small fraction of the terminal is invaded by an action potential or is capable of transmitter release when invaded, i.e., there are functional and nonfunctional boutons in the terminal axon. 1Vith repetitive stimulation, one-half the functional area could drop out. Alternatively, low quanta1 release could result from conduction block proximal to the whole terminal but with enough electrotonic current spread into the terminal to cause release from one or two sites. Functional boutons could release transmitter with a value of I) equal to or greater than the normal probability of release. This would esplain why the response doesn’t fail according to Poisson predictions. In fact, an estimate of p in the denervated extensor digitorum longus terminals, which was approximately four times higher than that in normal rat diaphragm, supports the interpretation that degenerating synapses have a greater probability of transmitter release from any individual release site but that the numbers of sites steadily decreases. REFERENCES 1. ALBUQUERQUE,
depolarization
E. X., F. T. SCFIUH, AND F. C. KAUFFMAN. 1971. Early membrane of the fast mammalian muscle after denervation. Pfliigcrs AYC!Z.
328 : 36-50. E., AND RAHAMIMOFF, R. 1975. On the role of mitochondria in transrelease from motor nerve terminals. J. Pltysiol. (Land.) 248: 285-306. BENNETT, M. R., E. Rf. MCLACHLAN, AXD R. S. TAI’LOR. 1973. The formation of synapses in reinnervated mammalian striated muscle. J. Pkysiol. (Lot~d.) 233: 481-500. BENNETT, M. R., T. FLORIN, AP\‘D R. Wooc. 1974. The formation of synapses in regenerating mammalian striated muscle. J. Pltysiol. (Lo&) 238 : 79-92. BOYD, I. A., AND A. R. MARTIN. 1956. The endplate potential in mammalian muscle. J. Physiol. (Lolzd.) 132: 74-91. BRAY, J. J., AR’D A. J. HARRIS. 1975. Dissociation between nerve-muscle transmission and nerve trophic effects on rat diaphragm using type D botulinum toxin.
2. ALP~AES,
mitter
3. 4. 5. 6.
J. Physiol. (Lorld.)
253 : 53-77.
488 7. 8.
9. 10.
11.
DIANA
J.
CARD
D. 1977. Denervation: Sequence of neuromuscular degenerative changes in rats and the effect of stimulation. Exp. Neural. 54: 251-265. CHRISTENSEN, B. N., AND A. R. MARTIN. 1970. Estimates of probability of transmitter release at the mammalian neuromuscular junction. J. Physiol. (Lo&.) 210 : 933-94s. DEL CASTILLO, J., AND B. KATZ. 1954. Quanta1 components of the endplate potential. J. Physiol. (Lond.) 124: 560-573. DESHPANDE, S. S., E. X. ALBUQUERQUE, AND L. GUTH. 1976. Neurotrophic regulation of prejunctional and postjunctional membrane at the mammalian motor endplate. Exp. Neural. 53 : 151-165. DUCHEN, L. W., AND E. STEFANI. 1971. Electrophysiological studies of neuromuscular transmission in hereditary “motor endplate disease” of the mouse. CARD,
J. Physiol.
(Lond.)
212:
535-548.
12. HUBBARD, J. I,, AND Y. L@YNING. 1966. The effects of hypoxia on neuromuscular transmission in a mammalian preparation. J. Physiol. (Lorcd.) 185: 205-223. 13. KRNJEVIC, K., AND R. MILEDI. 1958. Failure of neuromuscular propagation in rats. J. Physiol. (Lond.) 140: 440-461. 14. KRNJEVIC, K., AND R. MTLEDI. 1959. Presynaptic failure of neuromuscular propagation in rats. J. Physiol. (Lond.) 149: l-22. 15. KUNO, M. 1964. Quanta1 components of excitatory synaptic potentials in spinal motoneurons. J. Physiol. (Lond.) 175: 81-99. 16. MANOLOV, S. 1974. Initial changes in the neuromuscular synapses of denervated rat diaphragm. Brain Res. 65: 303-316. 17. MILEDI, R., AND C. R. SLATER. 1970. On the degeneration of rat neuromuscular junctions after nerve section. J. Physiol. (Load.) 207: 507-528. 18. WINLOW, W., AND D. N. R. USHERWOOD. 1975. Ultrastructural studies of normal and degenerating mouse neuromuscular junctions. J. Neurocytol. 4 : 377-394.
NEUROLOGY
Physiological Longus
54, 478-488 (1977)
Alterations Motor Nerve of Surgical
of Rat Extensor Digitorum Terminals as a Result Denervation
DIANA J. CARD 1 UnizIersity
of
Califontia,
Department Received
of Biology, July
Los Angeles,
California 90024
30,1976
The sequence of presynaptic degeneration changes which occur as a result of denervation of rat extensor digitorum longus muscle was studied. Reduction of endplate potential quanta1 content was the first neuromuscular alteration. Subsequently, endplate potentials disappeared although spontaneous transmitter release continued, at a higher frequency than in control muscles. During repetitive stimulation of the nerve, partial conduction block of the nerve terminal occurred at stimulation frequencies as low as I/s. Few terminals were able to respond at frequencies greater than 5/s. Transmitter release was quanta1 ; however, responses at synapses with a quanta1 content between one and five did not fit a Poisson distribution. The hypothesis is presented that, as the nerve terminal progressively degenerates after nerve transection, a few sites of transmitter release often persist for several hours, and these sites have a probability of release equal to or higher than normal.
INTRODUCTION Mammalian neuromuscular synapses are structurally and functionally changed as a result of denervation (1, 7, 10, 16-18). The first presynaptic alterations, resulting in increased miniature endplate potential (mepp) frequency several hours before spontaneous transmitter release disappears (1, 7, 18)) might be the result of depolarization of the nerve terminals (9) or they might involve the disruption of terminals and their engulfment by Schwann cells (16-18). These changescould be accompanied by alterations in coupling between invading presynaptic nerve action potentials and trans1 I would like to thank Dr. Alan Grinnell for his helpful advice and guidance throughout these experiments. This research was supported by USPHS Grant NS06232 to Dr. Grinnell and a predoctoral fellowship on USPHS Grant GM 448 to the author whose present address is Carnegie Institution of Washington, Department of Embryology, 115 West University Parkway, Baltimore, Maryland 21210. 478 Copyright All rights
@ 1977 by Academic Press, of reproduction in any form
Inc. reserved.
ISSN
0014-4886
MOTOR
NERVE
TERMINALS
AND
DENERVATION
479
mitter releaseor by failure of nerve action potentials to invade the terminals and could provide insight into the events occurring tluring degeneration. The purpose of the following experiments was to determine the sequence of presynaptic degenerative changes which occur as a result of nerve lesion. Other investigators have already documented postsynaptic alterations (1. 7, 10, 16-1s). The efficacy of evoked and spontaneous transmitter release was determined at different times after denervation. The occurrence of subthreshold endplate potentials, which is unusual in the absence of pre- or postsynaptic blocking agents, was then examined in detail. MATERIALS
AND METHODS
Extensor digitorum longus muscles from male Wistar rats (150 to 250 g) were denervated under light ether anesthesia using aseptic techniques. Surgical procedures and the methods for removal of the muscle, intracellular recording, and locating the neuromuscular junction were described previously (7). The contralateral extensor digitorum longus muscle served as a control: Its nerves were carefully separated from surrounding tissuesbut were not crushed or cut. Newowmm~lar Jmctio~ Recording. Spontaneous miniature endplate potentials with rise times of less than 1 ms were used as the criterion for locating neuromuscular junctions. They were observed for several minutes and either photographed on an oscilloscope screen or recorded with a Clevite-Brush Mark 220 pen recorder. Then the presence or absence of neuromuscular transmission at the synapse was tested by stimulating the entire extensor digitorum longus nerve stump using a suction electrode and recording the muscle fiber’s response. No attempt was made to reduce muscle twitching or endplate potential amplitude with curare or presynaptic blocking agents. The muscles were bathed in warmed (25 to 26”C), continuously perfused, oxygenated (95% 02, 5% CO,) bicarbonate-buffered saline (7) and remained in good condition for 8 h. No further denervation changes were observed after the muscles were placed in the bath; that is, synapsesat the same distance from the nerve cut gave similar responsesat the beginning and at the conclusion of the recording session. RESULTS Action Potentials. Until about 9 h after denervation, at endplates 1 mm from the nerve lesion, the only postsynaptic response to nerve stimulation was a muscle action potential (Table 1). The muscle twitched vigorously in responseto nerve activation. Muscle fibers in the control extensor digitorum longus always contracted in responseto nerve stimulation, even after 8 h in the bath.
480
DIANA
J.
TABLE Percentage
Distance (mm)
6-7
1
of Fibers in Denervated Extensor Digitorum Longus Exhibiting Action Potential, Endplate Potential, or No Responsea Denervation time (h)
1
CARD
Number fibers
Action potentials (%I
5 7 9 10 12 14
8.5 37 43 25 27 3
98 97 30 28 37 -
7 9 10 12 14 16 18 20
17 32 22 19 13 12 3 25
100 90 82 74 77 50 33 -
a All measurements were made a rise time of less than 1 ms.
from
fibers
having
Subthreshold endplate potentials (%I
an
No response (73
1
1 3 44 60 52 100
26 12 11 -
-
10
4 5 8 17 -
14 21 1.5 33 67 100
miniature
endplate
potentials
with
Subthreshold Endplate Potentials. Between 9 and 12 h after denervation, subthreshold endplate potentials were recorded from 5 to 20% of the synapses that had miniature endplate potentials (Fig. 1). The mean quanta1 content, WZ=
mean endplate potential amplitude mean miniature endplate potential amplitude’
Cl1
at these denervated neuromuscular junctions was greatly reduced compared to the response of the contralateral control and of denervated synapses prior to this time. Intermediate-size endplate potentials were observed whose amplitude varied in response to the same stimulus (Fig. la). At later denervation times the decrease in quanta1 content became so severe that small endplate potentials, with a quanta1 content between one and four, were often seen (Fig. lb). Four examples of this are summarized in Table 2. In these fibers, the endplate potential was usually the size of a single or a double miniature endplate potential, as opposed to the 100 to 200 quanta that comprise an endplate potential in normal muscle (5, 8). Subthreshold endplate potentials were never seen in contralateral control extensor digitorum longus muscles.
MOTOR
NERVE
TERMINALS
AiXD
481
DENERVATION
1mV
L
10 msec
2mV J 5msec
FIG. 1. a-Subthreshold endplate potential recorded from a synapse less than 1 mm from the cut nerve 8 h after denervation. Quanta1 content of this fiber was 17.5 ; miniature endplate potential frequency was 2.5/s. b-Low quanta1 endplate potential recorded at a synapse less than 1 mm from the nerve cut 8 h after denervation. Resting membrane potential was 83 mV. Quanta1 content was 1.2 ; miniature endplate potential frequency was 28.3/s.
After a 7-h denervation, the proportion of muscle fibers 1 mm from the nerve lesion with action potentials decreased, while the proportion of fibers with no response steadily increased. The percentage of synapseswith subthreshold endplate potentials was low but remained approximately conTABLE Quanta1
Fiber
A
Content of Four Different Extensor Digitorum
Resting potential
of failures
Actual
Predicted
84 111
0 46
31s
47 32 43
70 mV 1 /s 2/s
21-3
B
80 mv l/S
2/s
C
n
78 mv lis
21s ~1Measured
by Eqs. Cl] and
of
Muscle Fibers Longus Muscles”
Number
79 mv l/S
Number stimuli
2 from
Eq. Cl1
Eq. C31
Miniature endplate potentials per second
16 47
1.65 0.86
0.86
2.7
15 19 43
8 18
1.77 0.59 0
1.12 0.52
5.2
42 83
3 7
5 22
2.09 1.32
3.36 3.02
0.5
50 28
2 13
19 19
0.9i 0.39
3.22 0.77
3.8
[.~I at different
stimulation
frequencies.
Quanta1
Denervated
content
482
DIANA
J.
CARD
stant while synaptic transmission was present (Table 1). The synapses 6 to 7 mm from the nerve lesion showed the same pattern of deterioration of response, i.e., action potential, endplate potential, no response, as those at 1 mm ; however, the onset of each step occurred later (Table 1). The number of failures of subthreshold endplate potential response at low-frequency stimulation (l/s) in most cases, was fewer than predicted by Poisson’s law according to the quanta1 hypothesis. If x is the number of quanta in any one response, m is the mean of all values of x (the average of all responses of the number of quanta per response), and N is the total number of responses, then the frequency of endplate potentials, no/N, R/N, n2/N, in the classes x = 0, 1, 2, . . . can be determined by the Poisson distribution :
m” -nz = -pn* N x! The number of failures (when be predicted by the expression:
x = 0) at a given mean response
no/N
= ecm.
(+a) can
iI31
The average response (~2) was calculated by Eq. [ 11. At a synapse with quanta1 release, both ways of determining m, Eqs. [l] and [3] describe the mean quanta1 response (9). Therefore, 1n-N = n0
mean endplate potential amplitude mean miniature endplate potential amplitude’
c41
Table 2 shows the results of predictions of Poisson’s law compared with the actual number of failures when the nerve was stimulated. The actual number of failures did not agree with that predicted by Poisson. At l/s stimulation, the failure rate is lower than expected, except for fiber B. If the probability of quanta1 release is low (p approaches 0), then the frequency of endplate potential responses of x = 0, 1, 2, . . . approximates the Poisson distribution, and a large number of failures is expected. Because the number of failures of response at the degenerating synapses was lower than expected for the observed endplate potential amplitudes, the probability of transmitter release was estimated. Two binomial expressions : m = (1 - p)/v’ (where
ZJ= coefficient of variation
of endplate potential
-P m = In (1 - p)
lnN no’
responses)
and
MOTOR
NERVE
TERMINALS
AND
483
DENERVATION
were cc~rrhinrtl to give 1 P 2 = (1 - p) In (1 - p) ln?+ The quantity
l/v’
was plotted versus In no/N and the slope of the graph,
P (1 - P>In (1 - PI’ was calculated to give an estimate of p [Fig. 9 in ( 15) 1. At the degenerating extensor digitorum longus nerve terminals, p was estimated to be approximately 0.8 compared to that measured by other investigators in curarized rat diaphragm, about 0.2 (8). When the nerve was stimulated at frequencies above l/s, the number of failures increased, so that in some cases the actual and predicted number of failures was the same (Table 2, Fig. 2, 3). In Fig. 2A, the actual response amplitude remained the same, but the number of zero responses increased. In the muscle fiber of Fig. 2B, the endplate potential response size decreased along with an increased number of failures at higher stimulation frequencies. Figure 3 illustrates both reduced endplate potential size and failure of response in a muscle fiber stimulated at rates from one stimulus every 2 s to 10/s. Each graph represents the sequence of responses of the muscle fiber (endplate potentials) at the designated frequency. The muscle was allowed to rest for 1 min between each set of stimuli. Failure (no response) first 6.
J/set
mu 6AC6 o‘e ::I2 0
0.2
04
06
mV
0.8
ID
mV
FIG. 2. Endplate and miniature endplate potential (mepp) amplitude histograms. Endplate potentials were measured at a stimulation rate of 1, 2, or 3/s. These graphs illustrate that release is probably quanta1 and consists of only one, two, or three units. Increasing the stimulation rate increases the number of response failures. A-Fiber A of Table 2; B-Fiber B.
484
DIANA
J.
CARD
I/ 2sec
,;)&
io” ’ 40’“’ 63 ’ do* ’ IASsec
S/set
41JJ!Lec sequence
I om 20’” of responses
FIG. 3. Endplate potential size in millivolts of the lst, 2nd, 3rd, . . ., 20th, . . ., 40th, . . ., 60th, . . . etc. response when the nerve is stimulated every 2, 1.5, 1 s, etc. The muscle was allowed to rest 1 min between each stimulus regime.
appeared at a stimulation rate of 2/s. With increasing frequency of stimulation the number of failures also increased. Additionally, the number of full responsesat the beginning of the stimulation varied little with increasing stimulation frequency : A stepwise decrease of endplate potential amplitude occurred after about 10 stimuli at all frequencies. Because the responsewas decreased to less than half from stimulus 10 to 11 (at l/s), it is possible that the amount of nerve terminal releasing transmitter was similarly reduced after 10 stimuli. In fact, the initial “full” response of this terminal probably represents a vastly reduced releasearea compared to normal junctions. Some muscle fibers that spiked in responseto nerve stimulation occasionally failed to respond even with a subthreshold endplate potential at stimulation frequencies of 5 or 10/s, as if conduction block occurred at a branch point of the motor unit before the nerve terminal.
MOTOR
NERVE
TERMIXALS
AND
485
DENERVATION
0.p. L..
.
c zo:
3z
I ““I ePP
iotJ
f
II
A-“,.,,..,,..,.,...., zo-
no ePP
0
I” IO
7 mepps
I
20
r
I
30
T 1-1
40
/second
FIG. 4. Miniature endplate potential frequency versus frequency of occurrence in muscle fibers with a suprathreshold response eliciting action potentials (a.p.), subthreshold endplate potentials (epp) , and no response to indirect stimulation (no epp) . There are more fibers with higher frequencies in the “no epp.” category; however, the majority of values in each case is similar.
In summary, just prior to failure of synaptic transmission, the degenerating terminal was unable to follow stimuli at or greater than 2/s, resulting in depression of the endplate potential and possibly conduction block of the nerve action potential. The low quanta1 endplate potentials probably result from a reduced area of nerve terminal which releasestransmitter at a probability equal to or greater than normal. Disappearance of Endplate Potentials. Evoked release started disappearing about 9 h after denervation, although spontaneous transmitter release was still present at those synapses. Kot all fibers in the same region of the muscle became nonfunctional simultaneously. About the time evoked release disappeared, miniature endplate potential frequency increased (7). Then miniature endplate potentials disappeared and no synaptic response could be measured in the muscle. An attempt was made to correlate miniature endplate potential frequency with efficiency of evoked transmission. Figure 4 shows that fibers with no endplate potential have a broader range of miniature endplate potential frequencies than fibers with a subthreshold endplate potential or an action potential. However, most of the fibers measured in all groups have similar miniature endplate potential frequencies. DISCUSSION The sequence of presynaptic degenerative changes occurring at the extensor digitorum longus neuromuscular junction was : (i) reduced transmitter output and a subthreshold endplate potential, (ii) inability of the terminal to evoke synchronous transmitter release, although spontaneous miniature endplate potentials were still present or even increased in frequency, and (iii) disappearance of nliniature endplate potentials.
486
DIANA
J.
CARD
A permanent reduction of functional nerve terminal area as a result of denervation could account for the decrease in endplate potential quanta1 content observed after 9 h of denervation. As the nerve terminal degenerates, the amount of functional membrane may be progressively reduced so that the response size dwindles to nothing as fewer and fewer sites of transmitter release remain. Many degenerating axon terminals contain swollen and disrupted mitochondria (1618). Mitochondrial dysfunction could produce bursts of transmitter release (2, 12) and diminish or eliminate endplate potentials. As the nerve terminals become fragmented and engulfed by Schwann cells (17, 18) all synaptic activity disappears (17). Accordingly, when degenerating terminals are fixed in glutaraldehyde immediately after miniature endplate potentials disappeared, one sees numerous partially engulfed terminal axons lacking synaptic vesicles ( 18). Branch point blockade in the motor axon especially at high stimulation frequencies could explain the stepwise decline in response during a train. This would represent a reversible decrease in the area of terminal that is capable of releasing transmitter. After the spike was blocked in one portion of the nerve terminal, the other(s) continued to release quanta in a fluctuating manner, characteristic of endplate potentials recorded from Mg-blocked preparations. Complete failure of response could be due to conduction block proximal to the endplate, such as at a branch of the motor neuron to another muscle fiber of the motor unit. Normal mammalian neuromuscular junctions do exhibit conduction block, but the frequencies necessary to produce block are 20 to 40/s rather than the 2 to 5/s seen in this study (13, 14). However, the fact that the denervated nerve terminal is degenerating could probably account for the reduced conduction at branch points. With stimulation frequencies above l/s, the actual number of complete failures of response, in some cases, agrees with that predicted by Poisson statistics. The probability of release is probably decreased in these cases by failure of the action potential to invade the nerve terminals. Therefore, the similarity of the actual number of failures and those predicted by Poisson statistics may be coincidental. Subthreshold endplate potentials have been reported at regenerating neuromuscular junctions (3, 4). The sequence of establishment of transmission at the regenerating synapses is the reverse of that seen at degenerating endplates in this study: Miniature endplate potentials are usually seen before endplate potentials, then subthreshold endplate potentials occur, and finally muscle action potentials occur in response to nerve stimulation (3, 4). Subthreshold endplate potentials are also found in muscles partially blocked by botulinum toxin (6) and at neuromuscular junctions of mice affected by “motor endplate disease” (I 1). In each of these cases, the area of terminal capable of evoking synchronous transmitter release is probably
MOTOR
KERVE
TERMINALS
AND
DENERVATION
487
reduced. In particular, at synapses of the “motor endplate disease” mutant, action potentials fail to invade nerve terminals (11). Subthreshold endplate potentials have not been seen in degenerating rat diaphragm (17). Transmission loss occurs abruptly ; nerve stimulation either produces a muscle action potential or no response. The responses seen in the present study that have extremely low quanta1 content (e.g., wz = 2), but also a low failure rate, are unusual and are difficult to explain purely on the basis of the quanta1 hypothesis. Normally, for quanta1 analysis, increased magnesium and reduced calcium concentrations are used to decrease the probability of a particular quanta1 unit escaping from the nerve terminal (9). At normal neuromuscular junctions there is a relatively high probability of transmitter release and a correspondingly large surface area of functional terminal axon. One interpretation of the present data is that at this stage of degeneration only a small fraction of the terminal is invaded by an action potential or is capable of transmitter release when invaded, i.e., there are functional and nonfunctional boutons in the terminal axon. 1Vith repetitive stimulation, one-half the functional area could drop out. Alternatively, low quanta1 release could result from conduction block proximal to the whole terminal but with enough electrotonic current spread into the terminal to cause release from one or two sites. Functional boutons could release transmitter with a value of I) equal to or greater than the normal probability of release. This would esplain why the response doesn’t fail according to Poisson predictions. In fact, an estimate of p in the denervated extensor digitorum longus terminals, which was approximately four times higher than that in normal rat diaphragm, supports the interpretation that degenerating synapses have a greater probability of transmitter release from any individual release site but that the numbers of sites steadily decreases. REFERENCES 1. ALBUQUERQUE,
depolarization
E. X., F. T. SCFIUH, AND F. C. KAUFFMAN. 1971. Early membrane of the fast mammalian muscle after denervation. Pfliigcrs AYC!Z.
328 : 36-50. E., AND RAHAMIMOFF, R. 1975. On the role of mitochondria in transrelease from motor nerve terminals. J. Pltysiol. (Land.) 248: 285-306. BENNETT, M. R., E. Rf. MCLACHLAN, AXD R. S. TAI’LOR. 1973. The formation of synapses in reinnervated mammalian striated muscle. J. Pkysiol. (Lot~d.) 233: 481-500. BENNETT, M. R., T. FLORIN, AP\‘D R. Wooc. 1974. The formation of synapses in regenerating mammalian striated muscle. J. Pltysiol. (Lo&) 238 : 79-92. BOYD, I. A., AND A. R. MARTIN. 1956. The endplate potential in mammalian muscle. J. Physiol. (Lolzd.) 132: 74-91. BRAY, J. J., AR’D A. J. HARRIS. 1975. Dissociation between nerve-muscle transmission and nerve trophic effects on rat diaphragm using type D botulinum toxin.
2. ALP~AES,
mitter
3. 4. 5. 6.
J. Physiol. (Lorld.)
253 : 53-77.
488 7. 8.
9. 10.
11.
DIANA
J.
CARD
D. 1977. Denervation: Sequence of neuromuscular degenerative changes in rats and the effect of stimulation. Exp. Neural. 54: 251-265. CHRISTENSEN, B. N., AND A. R. MARTIN. 1970. Estimates of probability of transmitter release at the mammalian neuromuscular junction. J. Physiol. (Lo&.) 210 : 933-94s. DEL CASTILLO, J., AND B. KATZ. 1954. Quanta1 components of the endplate potential. J. Physiol. (Lond.) 124: 560-573. DESHPANDE, S. S., E. X. ALBUQUERQUE, AND L. GUTH. 1976. Neurotrophic regulation of prejunctional and postjunctional membrane at the mammalian motor endplate. Exp. Neural. 53 : 151-165. DUCHEN, L. W., AND E. STEFANI. 1971. Electrophysiological studies of neuromuscular transmission in hereditary “motor endplate disease” of the mouse. CARD,
J. Physiol.
(Lond.)
212:
535-548.
12. HUBBARD, J. I,, AND Y. L@YNING. 1966. The effects of hypoxia on neuromuscular transmission in a mammalian preparation. J. Physiol. (Lorcd.) 185: 205-223. 13. KRNJEVIC, K., AND R. MILEDI. 1958. Failure of neuromuscular propagation in rats. J. Physiol. (Lond.) 140: 440-461. 14. KRNJEVIC, K., AND R. MTLEDI. 1959. Presynaptic failure of neuromuscular propagation in rats. J. Physiol. (Lond.) 149: l-22. 15. KUNO, M. 1964. Quanta1 components of excitatory synaptic potentials in spinal motoneurons. J. Physiol. (Lond.) 175: 81-99. 16. MANOLOV, S. 1974. Initial changes in the neuromuscular synapses of denervated rat diaphragm. Brain Res. 65: 303-316. 17. MILEDI, R., AND C. R. SLATER. 1970. On the degeneration of rat neuromuscular junctions after nerve section. J. Physiol. (Load.) 207: 507-528. 18. WINLOW, W., AND D. N. R. USHERWOOD. 1975. Ultrastructural studies of normal and degenerating mouse neuromuscular junctions. J. Neurocytol. 4 : 377-394.
